#include <quda.h>
Definition at line 406 of file quda.h.
◆ a_max
double QudaEigParam_s::a_max |
◆ a_min
double QudaEigParam_s::a_min |
Range used in polynomial acceleration
Definition at line 425 of file quda.h.
◆ arpack_check
In the test function, cross check the device result against ARPACK
Definition at line 492 of file quda.h.
◆ arpack_logfile
char QudaEigParam_s::arpack_logfile[512] |
For Arpack cross check, name of the Arpack logfile
Definition at line 494 of file quda.h.
◆ batched_rotate
int QudaEigParam_s::batched_rotate |
For the Ritz rotation, the maximal number of extra vectors the solver may allocate
Definition at line 487 of file quda.h.
◆ block_size
int QudaEigParam_s::block_size |
For block method solvers, the block size
Definition at line 489 of file quda.h.
◆ check_interval
int QudaEigParam_s::check_interval |
For IRLM/IRAM, check every nth restart
Definition at line 483 of file quda.h.
◆ compute_gamma5
Performs the \gamma_5 OP solve by Post multipling the eignvectors with \gamma_5 before computing the eigenvalues
Definition at line 460 of file quda.h.
◆ compute_svd
Performs an MdagM solve, then constructs the left and right SVD.
Definition at line 456 of file quda.h.
◆ cuda_prec_ritz
The precision of the Ritz vectors
Definition at line 510 of file quda.h.
◆ eig_type
Type of eigensolver algorithm to employ
Definition at line 416 of file quda.h.
◆ extlib_type
Which external library to use in the deflation operations (MAGMA or Eigen)
Definition at line 542 of file quda.h.
◆ gflops
double QudaEigParam_s::gflops |
The Gflops rate of the eigensolver setup The time taken by the eigensolver setup
Definition at line 536 of file quda.h.
◆ import_vectors
Whether to load eigenvectors
Definition at line 507 of file quda.h.
◆ invert_param
Used to store information pertinent to the operator
Definition at line 413 of file quda.h.
◆ io_parity_inflate
Whether to inflate single-parity eigen-vector I/O to a full field (e.g., enabling this is required for compatability with MILC I/O)
Definition at line 533 of file quda.h.
◆ location
Location where deflation should be done
Definition at line 516 of file quda.h.
◆ max_restarts
int QudaEigParam_s::max_restarts |
For IRLM/IRAM, quit after n restarts
Definition at line 485 of file quda.h.
◆ mem_type_ritz
The memory type used to keep the Ritz vectors
Definition at line 513 of file quda.h.
◆ n_conv
int QudaEigParam_s::n_conv |
Number of requested converged eigenvectors
Definition at line 475 of file quda.h.
◆ n_ev
Size of the eigenvector search space
Definition at line 469 of file quda.h.
◆ n_ev_deflate
int QudaEigParam_s::n_ev_deflate |
Number of requested converged eigenvectors to use in deflation
Definition at line 477 of file quda.h.
◆ n_kr
Total size of Krylov space
Definition at line 471 of file quda.h.
◆ nk
◆ nLockedMax
int QudaEigParam_s::nLockedMax |
Max number of locked eigenpairs (deduced at runtime)
Definition at line 473 of file quda.h.
◆ np
◆ poly_deg
int QudaEigParam_s::poly_deg |
Degree of the Chebysev polynomial
Definition at line 422 of file quda.h.
◆ preserve_deflation
Whether to preserve the deflation space between solves. If true, the space will be stored in an instance of the deflation_space struct, pointed to by preserve_deflation_space
Definition at line 431 of file quda.h.
◆ preserve_deflation_space
void* QudaEigParam_s::preserve_deflation_space |
This is where we store the deflation space. This will point to an instance of deflation_space. When a deflated solver is enabled, the deflation space will be obtained from this.
Definition at line 435 of file quda.h.
◆ preserve_evals
If we restore the deflation space, this boolean indicates whether we are also preserving the evalues or recomputing them. For example if a different mass shift is being used than the one used to generate the space, then this should be false, but preserve_deflation would be true
Definition at line 442 of file quda.h.
◆ qr_tol
double QudaEigParam_s::qr_tol |
Tolerance on the QR iteration
Definition at line 481 of file quda.h.
◆ QUDA_logfile
char QudaEigParam_s::QUDA_logfile[512] |
Name of the QUDA logfile (residua, upper Hessenberg/tridiag matrix updates)
Definition at line 497 of file quda.h.
◆ require_convergence
If true, the solver will error out if the convergence criteria are not met
Definition at line 463 of file quda.h.
◆ run_verify
Whether to run the verification checks once set up is complete
Definition at line 519 of file quda.h.
◆ save_prec
The precision with which to save the vectors
Definition at line 528 of file quda.h.
◆ secs
double QudaEigParam_s::secs |
◆ spectrum
Which part of the spectrum to solve
Definition at line 466 of file quda.h.
◆ struct_size
size_t QudaEigParam_s::struct_size |
Size of this struct in bytes. Used to ensure that the host application and QUDA see the same struct size
Definition at line 408 of file quda.h.
◆ tol
double QudaEigParam_s::tol |
Tolerance on the least well known eigenvalue's residual
Definition at line 479 of file quda.h.
◆ use_dagger
What type of Dirac operator we are using If !(use_norm_op) && !(use_dagger) use M. If use_dagger, use Mdag If use_norm_op, use MdagM If use_norm_op && use_dagger use MMdag.
Definition at line 449 of file quda.h.
◆ use_eigen_qr
Use Eigen routines to eigensolve the upper Hessenberg via QR
Definition at line 453 of file quda.h.
◆ use_norm_op
◆ use_poly_acc
Use Polynomial Acceleration
Definition at line 419 of file quda.h.
◆ vec_infile
char QudaEigParam_s::vec_infile[256] |
Filename prefix where to load the null-space vectors
Definition at line 522 of file quda.h.
◆ vec_outfile
char QudaEigParam_s::vec_outfile[256] |
Filename prefix for where to save the null-space vectors
Definition at line 525 of file quda.h.
The documentation for this struct was generated from the following file: